ISO 5078:2025

Overview

ISO 5078:2025 - Management of terminology resources - Terminology extraction specifies methods for extracting candidate terms and terminological data from text corpora and gives guidance for selecting relevant designations, definitions, concept relations and other terminology-related information. The standard defines key concepts (e.g., candidate term, termhood, text corpus, terminological data) and provides a reference framework to improve terminology extraction tools and workflows.

Key topics and technical requirements

• Scope and definitions
  • Formal definitions for terms used in extraction workflows (candidate term, termhood, token, n‑gram, bitext, metadata, stop word).
• Text corpus compilation
  • Guidance on types of text corpora, selection criteria, and considerations for corpus creation to support reliable terminology extraction.
• Extraction approaches and methods
  • Classification of extraction approaches by language coverage, process, technique and technology. The document covers mainstream approaches (statistical, linguistic, hybrid and neural) without prescribing a single tool.
• Statistical and evaluative measures
  • Use of metrics such as TF‑IDF, keyness, precision, recall, and concepts of noise and silence to assess extraction outputs.
• Output handling and validation
  • Methods for filtering candidate term lists, assessing term eligibility, and reducing noise to maximize relevance.
• Workflow and implementation
  • End‑to‑end workflow stages: corpus selection/building, preprocessing, identifying candidate terms, selecting relevant terms, allocating terms to concepts, identifying concept relations and completing terminological entries.
• Tool characteristics
  • Guidance on tool features and expected behavior to optimize extraction performance and integration with terminology management systems.

Practical applications and users

ISO 5078:2025 is valuable for organizations and professionals who create and maintain terminology resources and language assets, including:

• Terminologists and language professionals building glossaries and termbases
• Ontology engineers extracting concepts and concept relations for knowledge models
• Information and data scientists using terminology for information retrieval, search optimization, and text analytics
• Localization and translation teams leveraging bitexts and term extraction to improve translation memories and machine translation quality
• Software vendors developing terminology extraction, NLP and knowledge‑management tools

Practical uses include populating terminology databases, accelerating glossary creation, improving term consistency across documents, and supporting ontology construction.

Related standards

• ISO 704 - Terminology work - Principles and methods
• ISO 1087 - Terminology work and terminology science - Vocabulary
• ISO 16642 - Terminological markup framework
• ISO 26162-1 - Terminology databases - Design

Keywords: ISO 5078, terminology extraction, terminology management, text corpora, candidate terms, term extraction, TF‑IDF, termhood, precision and recall, terminological data.